Ladder of Agency

The Permission Trap

“If only they would give me a chance” places the next opportunity in the hands of an investor, employer or institution. It treats agency as permission to act. The ladder of agency is Will's way of asking a different question: what demonstrated capability makes the next scale of action possible.

The levels describe what someone is doing with a situation: at first how much of it they can see, and from Level 4 upward how large a thing they are changing. Each level leans on the one before. Good intentions alone do not give a child dreaming of ending world hunger the ability to command civilizational resources. The missing capabilities lie between the intention and the scale of coordination it requires. Recognition or permission does not supply them, and a person handed the larger scale early still has the smaller capabilities to build.

A closer look

A change in the scale of agency

This is an orientation to the article's ten-level proposal, not a replacement for its individual levels.

Read this diagram

Levels in order: Assess a situation; Intervene in it; Coordinate with others; Build larger systems.

Level 1-3: From Binary Assessment to Probability Distributions

A startup provides a way to follow the first three levels through the same decision.

At Level 1, binary risk assessment, the startup is classified as risky or safe. “Startups are risky, most fail, therefore this is dangerous” supplies a judgment, but no operation that changes the risk. It is a static forecast.

At Level 2, causal models, the evaluator identifies what produces the risk: missing distribution channels, a crowded market and a team without prior exits. These explanations reveal causal mechanisms. The model is more informative, but the mechanisms are still treated as fixed conditions to describe.

At Level 3, probabilistic structure, risk becomes a distribution of possible paths and outcomes. Individual paths are microstates; aggregate outcomes are macrostates. Most startups fail, yet extreme successes can dominate expected value. A proposed distribution of 90% failure, 8% modest success and 2% outsized returns also invites the question of where those outcomes concentrate. Success is not uniform across the parameter space. This is the move toward statistical-mechanics reasoning rather than binary classification.

It is easy to stop at Level 2. Explaining a mechanism can feel like understanding the whole system even when the mechanism has not yet become an object of intervention.

Level 4: Engineering Probability Distributions

At Level 4, the generator of outcomes is changed. A forcing function reshapes the distribution of what can happen, rather than changing just one result.

For the startup, Level 3 might revise a 90% failure base rate to 70% after considering favorable signals. Level 4 asks which intervention would move the probability mass itself. A distribution agreement locked in before launch means the product no longer depends on finding a channel afterward. Prevention architecture takes selected failure modes out of play by removing the situations in which they occur. Consistent execution patterns compound momentum. Each of these changes the odds themselves.

The difference is comparable to forecasting rain versus engineering the climate. “Will this work?” evaluates a given arrangement. “What makes this work?” asks how to construct one. To someone treating risk as inherent and fixed, changing the distribution can look delusional or overconfident. The disagreement concerns whether the generator can be modified.

Levels 1–3 perform analysis; Level 4 performs engineering. Knowing why startups fail supplies knowledge. Building conditions that alter their chance of success supplies cybernetic control.

The Body as First Rung

Will puts the body first on the ladder: it is where Level 4 is shown to work before Level 5 brings in other people. The body is the system a person has the most say over. Changing it needs no one else's agreement and no one's permission, and the result is visible and cannot be obtained through verbal claims, credentials or networking.

That result also communicates before speech. A visibly changed body is evidence of deferring gratification, executing over months, recognizing actual results and changing a system's probabilities. It supplies an Alpha signal of demonstrated control. Physical state also underlies energy, cognition, emotional regulation and presence, so its effects extend to later work.

The body provides direct contact with an outcome: the transformation happened or it did not. Claiming the ability to build a company with $100M ARR while failing to manage nutrition and consistent training for 30 days creates a credibility deficit. The claim to coordinate fifty people conflicts with the absence of demonstrated coordination of one's own behavior.

That is why the ladder puts the body before teaching and coordination. A physical result is evidence of a kind that cleverness, articulation and connections do not provide. Someone asked to trust a person's control of a large external system has that person's handling of their own system as the nearest thing to check.

Level 5: Embodied Teaching and Coordination

Level 5 coordinates several agents toward a shared objective. A team, company or organization requires more than the founder's individual execution; it may involve ten, fifty or two hundred people acting in alignment.

What makes the teaching worth having is embodied use. Someone who has transformed their body through prevention, tracking and activation-cost management has encountered the details that another person will need. Reading the same methods without using them lacks that experience. The distinction between a generative source and a performance of expertise shows in those details: the first can answer from having done it, and the second can repeat the description.

Motivation affects whether the demonstrated capability stays in use. “I'll get jacked to attract a partner” contains a stopping condition: obtaining the partner completes the goal, and nothing in the goal then asks for the training to continue. “I'll transform my body as embodied proof that systematic thinking works” makes the accomplishment a step toward teaching. Reaching it enables further work instead of completing the mission.

A company built to realize a vision follows the same structure. Its purpose expands with capability. A company built to prove doubters wrong loses its motive when validation arrives or the doubters cease to matter. The expected value of continuing therefore changes differently in the two cases. Level 5 depends on a demonstrated capability that can be transmitted and sustained.

Level 6-10: Scaling Through Meta-Systems

Each later level applies earlier capability to a larger kind of system.

Level 6, meta-coordination, creates systems that create systems. A startup studio produces companies. Teachers train teachers who develop curricula. A pedagogy framework is applied in new domains, and a research methodology supports thousands of later researchers. These are probability-changing structures that generate further structures of the same kind.

Level 7, domain-general mastery, carries the same questions into other domains. Health, relationships, markets, cities and institutions can each be examined for forcing functions, states and feedback. The habit of looking for them is no longer confined to startups or fitness.

Level 8, institution creation, retains the framework beyond its operator. Universities, research institutions, movements and foundations continue developing it through changes of personnel. They are self-sustaining generators, allowing knowledge to compound across generations instead of disappearing when the founder leaves.

Level 9, civilizational reallocation, changes where capital and talent go. The surrounding field of activity changes, not merely one outcome within it. A demonstrated path makes a category of apparently unsolvable problems look tractable, and capital and talent then move toward it.

Level 10, ontological expansion, changes what people believe the species can do. Newton expanded what understanding nature meant; Einstein changed what counted as explanation; von Neumann established computation as a way of seeing. Their work altered the human conception of what could be achieved, moved or known.

Each level supplies evidence that the next one leans on. Transformation demonstrates mastery; coordination demonstrates transmission; meta-systems demonstrate abstraction; domain-generality demonstrates robustness; institutions demonstrate sustainability; civilizational change demonstrates impact; ontological expansion moves the perceived species-level ceiling. At every level the evidence is a result, and a claim does not stand in for one.

The Golden Orb: Authentic Generative Source

In Will's account, a framework that sustains this progression comes from the builder's own encounters. The authentic generative state, which he calls the golden orb, provides fuel that copying someone else's visible routines does not.

Observing someone who works at a much larger scale and imitating their habits preserves an external reference: “I should be like them.” The behavior still responds to someone else's surface rather than generating from the system at hand. “How would Steve Jobs handle this?” searches cached examples. “What does this system require?” starts a different search, directed at the current reality and able to produce a new solution.

The sun comparison concerns indirect effects. Solar energy heats the ocean, evaporation follows, pressure differences produce wind and leaves move. The sun does not persuade each leaf. Likewise, a generative source can change surrounding conditions without winning agreement from every skeptic.

Will's mechanistic framework developed through his constraints, reconstruction and debugging over time. The golden orb became systematic through contact with what was true of those systems. It is Alpha because it originates from those encounters rather than from another person's model. Copying a successful person's surface is the corresponding Beta behavior: responsive imitation presented as independent ambition.

Someone building because they see what could exist has a vision that grows with their capability. Someone building for external markers exhausts the motive once they receive them. The first form of intrinsic motivation still has something to do after each success, which matters for work on the scale of decades that Level 8 and beyond would take.

The Motivation Architecture: Purpose Versus Validation

The stopping condition can be followed through several examples. “I'll get jacked to prove my ex wrong” ends when the ex sees the result and supplies validation. Fitness then loses its purpose, and nothing is left to keep the training going. “I'll build this company to show my dad I'm not a failure” ends when an exit produces the father's acknowledgment. The company then feels empty, because it was a means of obtaining that response.

By contrast, “I'll transform my body as proof that systematic thinking works, which unlocks my ability to teach these frameworks to others” makes the outcome usable in a further pursuit. The result is where the teaching at Level 5 starts from. “My body is proof the framework works” therefore has different motivational dynamics from “my body will make people like me,” even when the outward behavior is identical.

The same distinction applies to building a company to “prove I can” versus building because “I see what should exist and must instantiate it.” Research done to “get tenure” has a different endpoint from “this question demands answer and I must reveal it.” Teaching to “be recognized as expert” ends at recognition, while “these insights must propagate” extends through further transmission.

Validation sets a ceiling at the response being sought. Purpose expands as capability increases, so accomplishment supplies additional fuel. The institution-building and civilizational work beyond Level 6 is measured in decades. A motive that is complete when the sought response arrives will, if the response comes, be complete long before that work is.

Practical Implications: Using the Ladder

What the order protects against shows when a level is skipped. Teaching at Level 5 without having used the method at Level 4 leaves the teacher without the details a learner will ask for. Creating a Level 8 institution without Level 5 experience of coordination means learning coordination at the scale where its failures cost the most. Attempting Level 9 reallocation without Level 7 generality confines the impact to a single domain instead of translating it to adjacent ones.

Tracking and journaling retain the evidence for locating a current capability. “I understand distributions but am not yet engineering them consistently” places the current work between Level 3 and Level 4 and identifies what comes next. Claiming Level 6 while Level 4 is incomplete creates a false account of the current state that prevents learning.

On Will's ladder the first result is control of the body through consistent execution over enough time. That result is what the next attempt, at coordination, builds on. Each later level repeats the pattern: a capability is demonstrated at one scale before another system is asked to rely on it. An authentic framework supplies the continuing motive for doing that work.

Intelligence Design applies Level 4 distribution engineering to AI systems.
Agency examines the intent–execution interface and freedom at the microstate level.
Golden Orb distinguishes a generative source from copied surfaces.
Probability Space Bending changes distributions through intervention.
30x30 Pattern describes a month of daily repetition for making execution routine.
Prevention Architecture makes selected failures unavailable.
Cybernetics supplies feedback and control at each scale.
Expected Value connects probability with sustained motivation.
Forcing Functions changes the distribution of possible actions.
Activation Energy examines the thresholds crossed at Levels 1–4.
Signal Theory separates generative and responsive signals.
Question Theory traces the searches initiated by different questions.
Tracking records observable evidence of capability.
Journaling retains what each attempt teaches.
Paths and Trailblazing distinguishes a personally developed path from imitation.
Be the Sun places collection systems around a generative source at the top of the ladder.